Micro-grid energy storage device energy optimal management method

Abstract

The present application relates to the technical field of micro-grid, in particular to a kind of micro-grid energy storage equipment energy optimal management method, mainly solve the technical problem of lack of the method capable of optimal management of micro-grid energy in prior art.The present application provides a kind of micro-grid energy storage equipment energy optimal management method, which comprises the following steps: S1, establishing micro-grid system model;S2, constructing the performance index function of micro-grid energy management;S3, optimal performance index function and optimal control strategy are solved using iterative approximate dynamic programming algorithm, to obtain optimal management approach.The present application designs an iterative approximate dynamic programming algorithm to obtain optimal management approach, to realize the optimal management of micro-grid energy.

Description

Technical Field

[0001] This invention relates to the field of microgrid technology, and in particular to a method for optimal energy management of microgrid energy storage devices. Background Technology

[0002] In recent years, the development of renewable energy has created a new landscape for power generation and distribution. Microgrids, with their ability to integrate renewable energy sources such as wind and solar power, and their ability to operate independently during main grid outages, have become a key solution for improving energy efficiency, reliability, and sustainability. Against this backdrop of continuous development, and given the potential of microgrids in improving energy efficiency and reliability, research on microgrid energy management is crucial.

[0003] However, current research on microgrid energy management is still in its developmental stage, and there is no mature method for optimal energy management in microgrids. Therefore, there is an urgent need for an optimal energy management method for microgrid energy storage devices. Summary of the Invention

[0004] To overcome the technical deficiency of existing technologies that lack methods for optimal energy management of microgrids, this invention provides a method for optimal energy management of microgrid energy storage devices.

[0005] The energy optimization management method for microgrid energy storage devices provided by this invention includes the following steps:

[0006] S1. Establish a microgrid system model, including the main grid, loads, trams, photovoltaic modules, and batteries. The total power of the main grid, photovoltaic modules, and batteries must meet the needs of the loads and trams, satisfying the energy balance equation:

[0007] ;

[0008] in, For battery charging and discharging power, Main grid power, For photovoltaic power, For load power, For the power of the tram;

[0009] The battery model is as follows:

[0010] ;

[0011] in, for Battery level at all times Improve battery charging and discharging efficiency;

[0012] Define the system state as:

[0013] ;

[0014] in, ;

[0015] Control strategy ;

[0016] Therefore, the expression for the microgrid system model is:

[0017] ;

[0018] S2. Construct a performance index function for microgrid energy management, the expression of which is:

[0019] ;

[0020] in, The first item The second item refers to the costs or profits generated from transactions between the microgrid and the main grid. The goal is to bring energy storage closer to the center of its charge, the third item The purpose is to avoid excessive charging and discharging power. , and These are weight parameters;

[0021] According to Bellman's optimality principle, the expressions for the optimal performance index function and the optimal control strategy are:

[0022] ;

[0023] in, The optimal performance index function. The optimal control strategy;

[0024] S3. Use an iterative approximate dynamic programming algorithm to solve for the optimal performance index function and the optimal control strategy to obtain the optimal management method:

[0025] Construct a value function to solve for the optimal performance index function. The expression of the value function is:

[0026] ;

[0027] in, Iterative index;

[0028] Therefore, the expression for the battery control strategy is:

[0029] ;

[0030] along with By increasing the value of , the optimal performance index function and the optimal control strategy can be obtained.

[0031] Optionally, an evaluation neural network can be used to approximate the value function, and an action neural network can be used to approximate the control strategy.

[0032] Optionally, will The optimal value function at time t is expressed as:

[0033] ;

[0034] in, For ideal output weight, These are the basis vectors for the output. As input to the hidden layer, Network error is a value function;

[0035] The approximate value function is:

[0036] ;

[0037] in, In order to be in The first hour The weight parameters of the hidden layer in the next iteration It is a sigmoid function, the input of the hidden layer. for:

[0038] ;

[0039] Among them, the weights of the input layer It is fixed as a constant in the range [-1, 1]. It is the input of the input layer;

[0040] The expression for evaluating the iterative error of a neural network is:

[0041] ;

[0042] definition , ;

[0043] The expression for evaluating the iterative error of a neural network can then be rewritten as:

[0044] ;

[0045] Define auxiliary matrix and for:

[0046] ;

[0047] in, The weight sequence of historical iteration information. This is the forgetting factor, and its value ranges from (0,1).

[0048] and The iterative form is:

[0049] ;

[0050] The formula for updating the weight parameters of the hidden layer is:

[0051] ;

[0052] in, The time-varying gain matrix, This indicates that the weighted sum using historical iteration information is updated using the following formula:

[0053] .

[0054] Optionally, will The optimal control strategy at time t is expressed as:

[0055] ;

[0056] in, These are the optimal weight parameters for the action network. It is the sigmoid function;

[0057] The approximate control strategy is:

[0058] ;

[0059] in, The input to the hidden layer is an approximate weight. It can be represented as:

[0060] ;

[0061] in, A fixed weight vector ranging from -1 to 1;

[0062] The iterative error of the action neural network is:

[0063] ;

[0064] in, Let i and k represent the total cost objectives of all sums.

[0065] The formula for updating the weight parameters of an action neural network is:

[0066] ;

[0067] Learning rate It is set based on experience.

[0068] The technical solution provided by this invention has the following advantages compared with the prior art:

[0069] The present invention provides a method for optimal energy management of microgrid energy storage devices, which designs an iterative approximate dynamic programming algorithm to obtain the optimal management method, so as to achieve optimal energy management of microgrids. Attached Figure Description

[0070] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0071] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0072] Figure 1 This diagram illustrates the structure of the microgrid in an embodiment of the present invention.

[0073] Figure 2 This diagram illustrates the distribution of solar power over 168 hours in an embodiment of the present invention.

[0074] Figure 3 This represents a load distribution diagram of households over 168 hours in an embodiment of the present invention.

[0075] Figure 4 This diagram illustrates the power consumption distribution of an electric vehicle over 168 hours in an embodiment of the present invention.

[0076] Figure 5 This diagram illustrates the structure of the approximate dynamic programming algorithm in an embodiment of the present invention.

[0077] Figure 6 This represents an electricity price chart over a 168-hour period as described in this embodiment of the invention.

[0078] Figure 7 (a) shows the power grid learning performance diagram at the 95th hour in this embodiment of the invention;

[0079] Figure 7 (b) shows the learning performance graph of the battery control strategy at the 95th hour in this embodiment of the invention;

[0080] Figure 8 This diagram illustrates the main power grid power performance in an embodiment of the present invention.

[0081] Figure 9This diagram illustrates the battery control strategy in an embodiment of the present invention.

[0082] Figure 10 This diagram illustrates the battery energy performance in an embodiment of the present invention.

[0083] Figure 11 This diagram shows a comparison of the gradient descent method at the 95th hour in this embodiment of the invention with the proposed adaptive law on the value function. Detailed Implementation

[0084] To better understand the above-mentioned objectives, features, and advantages of the present invention, the solutions of the present invention will be further described below. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other.

[0085] Many specific details are set forth in the following description in order to provide a full understanding of the invention, but the invention may also be practiced in other ways different from those described herein; obviously, the embodiments in the specification are only some embodiments of the invention, and not all embodiments.

[0086] The following is combined Figures 1 to 11 Specific embodiments of the present invention will be described in detail below.

[0087] This embodiment provides a method for optimal energy management of microgrid energy storage devices, including steps S1 to S3.

[0088] S1. Establish a microgrid system model, including the main grid, loads, trams, photovoltaic modules, and batteries. The total power of the main grid, photovoltaic modules, and batteries must meet the needs of the loads and trams, satisfying the energy balance equation:

[0089] ;

[0090] in, For battery charging and discharging power, Main grid power, For photovoltaic power, For load power, For the power of the tram;

[0091] The battery model is as follows:

[0092] ;

[0093] in, for Battery level at all times Improve battery charging and discharging efficiency;

[0094] Define the system state as:

[0095] ;

[0096] in, ;

[0097] Control strategy ;

[0098] Therefore, the expression for the microgrid system model is:

[0099] .

[0100] It is easy to understand that there are maximum and minimum limits on battery charging and discharging power, photovoltaic power, and electric vehicle power, as follows:

[0101] , , .

[0102] Specifically, the formula for calculating battery charge and discharge efficiency is as follows:

[0103] .

[0104] S2. Construct a performance index function for microgrid energy management, the expression of which is:

[0105] ;

[0106] in, The first item The second item refers to the costs or profits generated from transactions between the microgrid and the main grid. The goal is to bring energy storage closer to the center of its charge, the third item The purpose is to avoid excessive charging and discharging power. , and These are weight parameters;

[0107] According to Bellman's optimality principle, the expressions for the optimal performance index function and the optimal control strategy are:

[0108] ;

[0109] in, The optimal performance index function. This is the optimal control strategy.

[0110] As is easily understood, this embodiment uses a multi-objective cost function as the performance index function, which can reduce electricity purchase costs and extend battery life.

[0111] It should be noted that, and This combination helps protect the battery and extend its lifespan.

[0112] It is easy to understand that by adjusting , and It can adapt to different needs.

[0113] S3. Use an iterative approximate dynamic programming algorithm to solve for the optimal performance index function and the optimal control strategy to obtain the optimal management method:

[0114] Construct a value function to solve for the optimal performance index function. The expression of the value function is:

[0115] ;

[0116] in, Iterative index;

[0117] Therefore, the expression for the battery control strategy is:

[0118] ;

[0119] along with By increasing the value of , the optimal performance index function and the optimal control strategy can be obtained.

[0120] It should be noted that the expressions for the optimal performance index function and the optimal control strategy obtained according to Bellman's optimality principle are difficult to solve directly. Therefore, this embodiment uses an iterative approximate dynamic programming algorithm to solve for the optimal performance index function and the optimal control strategy.

[0121] Specifically, in order to facilitate the implementation of the iterative approximate dynamic programming algorithm, this embodiment uses an evaluation neural network to approximate the value function and an action neural network to approximate the control strategy.

[0122] More specifically, the steps for evaluating the approximate value function of the neural network are as follows:

[0123] Will The optimal value function at time t is expressed as:

[0124] ;

[0125] in, For ideal output weight, These are the basis vectors for the output. As input to the hidden layer, Network error is a value function;

[0126] The approximate value function is:

[0127] ;

[0128] in, In order to be in The first hour The weight parameters of the hidden layer in the next iteration It is a sigmoid function, the input of the hidden layer. for:

[0129] ;

[0130] Among them, the weights of the input layer It is fixed as a constant in the range [-1, 1]. It is the input of the input layer;

[0131] The expression for evaluating the iterative error of a neural network is:

[0132] ;

[0133] definition , ;

[0134] The expression for evaluating the iterative error of a neural network can then be rewritten as:

[0135] ;

[0136] Define auxiliary matrix and for:

[0137] ;

[0138] in, The weight sequence of historical iteration information. This is the forgetting factor, and its value ranges from (0,1).

[0139] and The iterative form is:

[0140] ;

[0141] The formula for updating the weight parameters of the hidden layer is:

[0142] ;

[0143] in, The time-varying gain matrix, This indicates that the weighted sum using historical iteration information is updated using the following formula:

[0144] .

[0145] It should be noted that this embodiment uses a new discrete adaptive law to update the weight parameters of the evaluation neural network, which can accelerate the convergence speed and alleviate the risk of getting trapped in local optima, ensuring that the value function successfully approximates the optimal performance index function.

[0146] It should be noted that this embodiment establishes a new composite performance index function for the energy management system in a microgrid, which can minimize the cost of the microgrid and extend battery life.

[0147] More specifically, the steps for approximating the control strategy using an action neural network are as follows:

[0148] Will The optimal control strategy at time t is expressed as:

[0149] ;

[0150] in, These are the optimal weight parameters for the action network. It is the sigmoid function;

[0151] The approximate control strategy is:

[0152] ;

[0153] in, The input to the hidden layer is an approximate weight. It can be represented as:

[0154] ;

[0155] in, A fixed weight vector ranging from -1 to 1;

[0156] The iterative error of the action neural network is:

[0157] ;

[0158] in, Let i and k represent the total cost objectives of all sums.

[0159] The formula for updating the weight parameters of an action neural network is:

[0160] ;

[0161] Learning rate It is set based on experience.

[0162] In this embodiment, the battery capacity in the microgrid is 100 kWh. To extend its lifespan, the usable capacity is limited to the range of 20-80 kWh. The charge / discharge rate is limited to between -16 kW and 16 kW. The initial battery energy is set to 75 kWh. The weighting parameters in the performance index function are set as follows. , , The initial values ​​of the time-varying gain matrix and auxiliary matrix are set to... , , The forgetting factor is set to , , , and Set between [−1, 1].

[0163] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the present invention. Although detailed descriptions have been provided with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments, and they should all be covered within the protection scope of the claims.

Claims

1. A method for optimal energy management of microgrid energy storage devices, characterized in that, Includes the following steps: S1. Establish a microgrid system model, including the main grid, loads, trams, photovoltaic modules, and batteries. The total power of the main grid, photovoltaic modules, and batteries must meet the needs of the loads and trams, satisfying the energy balance equation: ； in, For battery charging and discharging power, Main grid power, For photovoltaic power, For load power, For the power of the tram; The battery model is as follows: ； in, for Battery level at all times Improve battery charging and discharging efficiency; Define the system state as: ； in, ; Control strategy ; Therefore, the expression for the microgrid system model is: ； S2. Construct a performance index function for microgrid energy management, the expression of which is: ； in, For performance index functions, The first item The second item refers to the costs or profits generated from transactions between the microgrid and the main grid. The goal is to bring energy storage closer to the center of its charge, the third item The purpose is to avoid excessive charging and discharging power. , and These are weight parameters; According to Bellman's optimality principle, the expressions for the optimal performance index function and the optimal control strategy are: ； in, The optimal performance index function. The optimal control strategy; S3. Use an iterative approximate dynamic programming algorithm to solve for the optimal performance index function and the optimal control strategy to obtain the optimal management method: Construct a value function to solve for the optimal performance index function. The expression of the value function is: ； in, It is a value function. For iterative indexing; Therefore, the expression for the battery control strategy is: ； For battery control strategies; along with By increasing the value of , the optimal performance index function and the optimal control strategy can be obtained; An evaluation neural network is used to approximate the value function, and an action neural network is used to approximate the control strategy. Will The optimal value function at time t is expressed as: ； in, for The optimal value function at time t. For ideal output weights, These are the basis vectors for the output. As input to the hidden layer, Network error is a value function; The approximate value function is: ； in, It is an approximate value function. In order to be in The first hour The weight parameters of the hidden layer in the next iteration It is a sigmoid function, the input of the hidden layer. for: ； Among them, the weights of the input layer It is fixed as a constant in the range [-1, 1]. It is the input of the input layer; The expression for evaluating the iterative error of a neural network is: ； definition , ; The expression for evaluating the iterative error of a neural network can then be rewritten as: ； Define auxiliary matrix and for: ； in, The weight sequence of historical iteration information. This is the forgetting factor, and its value ranges from (0,1). and The iterative form is: ； The formula for updating the weight parameters of the hidden layer is: ； in, These are the weight parameters of the hidden layer. The time-varying gain matrix, This indicates that the weighted sum using historical iteration information is updated using the following formula: 。 2. The energy optimization management method for microgrid energy storage devices according to claim 1, characterized in that, Will The optimal control strategy at time t is expressed as: ； in, for The optimal control strategy at any given time. These are the optimal weight parameters for the action network. It is the sigmoid function; The approximate control strategy is: ； in, For approximate control strategies, The input to the hidden layer is an approximate weight. It can be represented as: ； in, A fixed weight vector ranging from -1 to 1; The iterative error of the action neural network is: ； in, Indicates the overall cost target; The formula for updating the weight parameters of an action neural network is: ； in, The weight parameters and learning rate of the action neural network. It is set based on experience.

Images